Application of Deep Convolution Network Compression Algorithm in Weld Recognition

Type of layers	Parameter
Convolution+activation function+pooling	C₆₄(3×3,S=1)+RELU+P_max(2×2,S=2)
Convolution+activation function+pooling	C₁₂₈(3×3,S=1)+RELU+P_max(2×2,S=2)
Convolution+activation function+pooling	C₂₅₆(3×3,S=1)+RELU+P_avg(2×2,S=2)
Convolution+activation function+pooling	C₅₁₂(3×3,S=1)+RELU+P_avg(2×2,S=2)
Convolution+activation function+pooling	C₅₁₂(3×3,S=1)+RELU+P_avg(2×2,S=2)
Full connetion+activation function	400 neutral units+RELU
Full connetion+activation function	400 neutral units+RELU
Loss function	Euclidean distance

Table 2. Model size and memory requirements
View table
Table 2. Model size and memory requirements
Without compression After compression Compressionratio Speed ratio Loss /%
Maximum memoryrequirement /MB Modelsize /MB Maximum memoryrequirement /MB Modelsize /MB
38 55.3 1.1 2.33 23.8 26.3 1.4

Table 3. Test accuracy of weldment 1
View table
Table 3. Test accuracy of weldment 1
Method Accuracy /%
GS 94.3
SURF 92.4
HC 90.1
LSE 87.6
GH 93.3
GXN 97.5

Table 4. Test results of multiple weldments
View table
Table 4. Test results of multiple weldments
Method Accuracy /%
Secondweldment test Mixeddata test
GS 93.3 90.1
SURF 94.4 89.6
HC 87.5 86.7
LSE 94.3 88.2
GH 91.2 90.1
GXN 96.7 96.3

Table 5. Time consumption of each method
View table
Table 5. Time consumption of each method
Method Time /ms
GS 35
SURF 27
HC 17
LSE 39
GH 37
GXN 36

Tools

Get Citation

Copy Citation Text

Meiju Liu, Bo Yun. Application of Deep Convolution Network Compression Algorithm in Weld Recognition[J]. Laser & Optoelectronics Progress, 2019, 56(5): 051004

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites